arXiv:2605.23917v1 Announce Type: new Abstract: Modern scientific discovery is bottlenecked not by data scarcity, but by the inability to synthesize fragmented knowledge into actionable hypotheses. This challenge is especially acute in battery materials research, where electrochemical performance, interfacial behavior, and manufacturing feasibility must be optimized simultaneously. Here, we present the Multi-Persona Debate System (MPDS), a literature-grounded framework for automated scientific hypothesis generation that combines literature retrieval, long-context large language model reasoning, corpus-driven persona induction, and structured multi-agent debate. MPDS constructs literature snapshots of up to 500 papers, grounds agents in role-specific evidence pools, and conducts a three-round citation-aware debate followed by moderator synthesis, enabling negotiation between personas while preserving evidence traceability. We evaluate MPDS using a temporally controlled protocol excluding direct access to target papers, including two held-out battery-materials case studies and a blinded comparison across 30 matched cases. In sodium-ion anode and all-solid-state battery cathode design tasks, MPDS recovered design logics aligned with experimentally validated solution spaces and generated more mechanistically explicit, process-aware proposals than simpler baselines. To assess the impact of personas and debate, we introduce Integrative Hypothesis Quality scoring. In ablation studies, MPDS achieved the highest mean score among five conditions, with its largest advantage in cross-perspective integration. A laboratory follow-up suggests utility as a diagnostic aid for identifying practical bottlenecks in workflows. These results indicate that structured debate over literature snapshots improves hypothesis formation under coupled engineering constraints and provides a reusable workflow for text-intensive scientific discovery.

arXiv:2605.23997v1 Announce Type: new Abstract: Multimodal large language models via reinforcement learning (RL) have demonstrated remarkable capabilities in complex visual reasoning tasks, yet they remain limited in long-horizon multimodal scenarios, often suffering from visual hallucination and logical error. Current methods typically pre-encode high-dimensional visual scenes into discrete textual proxies to facilitate downstream reasoning. As the reasoning chain unfolds, however, the inherent information asymmetry between text and visual scenes tends to erode visual grounding, resulting in misguided reasoning and erroneous outputs. To address this issue, we introduce IVR-R1 (Iterative Visual-grounded Reasoning), a novel RL training framework that facilitates dynamic visual re-alignment that actively rectifies reasoning trajectories to guide policy optimization. Specifically, by leveraging a reward-driven screening mechanism to identify flawed rollouts, IVR-R1 executes a fine-grained, step-level error attribution within the multimodal context. By iteratively cross-referencing intermediate reasoning states against pristine visual priors, a Re-Reasoning Loop enables automated trajectory rectification, effectively synthesizing expert-level demonstrations that serve as high-fidelity reasoning templates for the policy model. Our experiments across diverse multimodal benchmarks demonstrate that IVR-R1 consistently outperforms existing reinforcement learning methods, establishing a superior paradigm for maintaining logical and visual consistency in complex multimodal reasoning.

arXiv:2605.24008v1 Announce Type: new Abstract: Fault detection for Deep Neural Networks (DNNs) has received increasing attention in recent years. While more advanced hybrid approaches have been proposed to combine multiple sources of information and outperform earlier techniques, they often incur substantial computational overhead, limiting scalability and practicality in real-world settings. In this paper, we introduce Concept-Aware Fault Detection (CAFD), a learning-based approach that achieves superior fault detection performance by effectively integrating multiple information sources while maintaining practical efficiency. Specifically, CAFD is trained using a carefully selected set of informative features, including model-based signals derived from the DNN's outputs, distance-based features, and a novel concept-based feature, called Concept Failure Ratio (CFR). CFR leverages Vision-Language Models (VLMs) to extract textual concepts from images and quantify the likelihood that their presence is associated with DNN failures. By incorporating this feature, CAFD benefits from complementary semantic information, enabling more effective fault detection. Our results demonstrate that CFR serves as an effective indicator for DNN fault detection. We conduct an extensive empirical evaluation of CAFD, comparing it against five state-of-the-art baselines across three subject DNN models and datasets, including ImageNet. Across a wide range of constrained selection budgets, CAFD consistently outperforms all baselines in Fault Detection Rate (FDR), achieving average FDR improvements of 18.3% across all investigated subjects and budget sizes.

Discussion | Link

<blockquote cite="https://twitter.com/quinnypig/status/2058960462256210268"><p>I cannot believe I'm saying this, but getting the literal Pope to canonize your product's specific technical limitations as a spiritual treatise is the single greatest act of vendor lobbying I have ever seen.</p></blockquote> <p class="cite">&mdash; <a href="https://twitter.com/quinnypig/status/2058960462256210268">Corey Quinn</a>, on Anthropic co-founder Christopher Olah's <a href="https://www.washingtonpost.com/world/2026/05/25/pope-elevates-ai-ethics-religious-imperative-with-first-encyclical/">influence</a> on <em>Magnifica Humanitas</em></p> <p>Tags: <a href="https://simonwillison.net/tags/ai-ethics">ai-ethics</a>, <a href="https://simonwillison.net/tags/corey-quinn">corey-quinn</a>, <a href="https://simonwillison.net/tags/anthropic">anthropic</a>, <a href="https://simonwillison.net/tags/ai">ai</a></p>

Discussion | Link

New York Times: A surge in AI-generated “pro se” cases, or lawsuits filed by self-represented litigants, is democratizing the legal system but consuming more court resources  —  For years, courts have welcomed cases brought by self-represented litigants.  Now those plaintiffs have A.I. …

Jack Topalian as Naz and Sydney Sweeney as Cassie in "Euphoria" season three. Patrick Wymore/HBO Warning: Spoilers ahead for "Euphoria" season three, episode seven, "Rain or Shine." Nate Jacobs (Jacob Elordi) met his gruesome end in "Euphoria" after being buried alive by mobsters. Showrunner Sam Levinson said the "horrific" scene was designed to make fans feel uncomfortable. HBO's once-beloved teen drama "Euphoria" has transformed into a violent, Quentin Tarantino-style crime thriller in its third season — a mutation now complete with a gruesome death for one of its main characters. In the season's seventh and penultimate episode, titled "Rain or Shine," Nate Jacobs (Jacob Elordi) is buried alive in a wooden coffin (hello, "Kill Bill") as revenge for his outstanding debt to Naz (Jack Topalian), a local businessman and mob-like figure. Nate's wife, Cassie Howard (Sydney Sweeney), is forced to come up with $1 million in 72 hours, lest Nate dies underground of dehydration. However, before she can succeed or fail, a rattlesnake slithers through the coffin's airhole. Once Cassie does get a chance to dig up her husband, she finds Nate's body purplish, bloated, and bloodshot, the poisonous snake coiled around it. The final shot of Nate's face could be ripped straight from a horror film. Nate has been a principal focus of "Euphoria" since its premiere in 2019. He was introduced as a manipulative high-school athlete with wells of pent-up rage; in season one alone, Nate choked his girlfriend, blackmailed a classmate, and plotted to send an innocent man to prison. He got away with it all. Season three features a five-year time jump, thrusting the show's ensemble into the harsh light of young adulthood. Many fans have been eager to see Nate finally pay for his crimes — a fact the show's creator and showrunner, Sam Levinson, said was top of mind while plotting the character's fate. "There's this kind of funny thing where I know what the audience wants in terms of justice or karma and with that in mind, I always think, 'Well, how can I give it to them?" Levinson told Esquire. "How can I give them what they want, but make it so horrific and anxiety-inducing that by the time it happens, the audience isn't so sure they wanted it?" Nate Jacobs (Jacob Elordi) is punished by mob-style criminals in "Euphoria" season three. Patrick Wymore/HBO Levinson said he hoped the audience would feel complicit in Nate's suffering this season, which also involved having multiple body parts cut off in his conflict with Naz. "It's like, 'Oh, you wanted him to get his comeuppance…? OK," Levinson continued. "You end up going, 'Oh God, I don't know. Should he have had it better? Did he deserve it?' Those kinds of questions are always exciting to pose to the audience." Levinson has been both praised and criticized for relying on shock value and rage bait in his work, especially in "Euphoria," which delights in portraying its young characters — especially the women — in vulnerable, compromising positions, both literal and figurative. (Sweeney, who's had the most nude scenes of her costars, has defended Levinson from backlash, saying the goal of "Euphoria" is "making people uncomfortable.") Levinson also prefers to use practical effects to achieve his desired results. He revealed to Esquire that while filming Nate's death sequence, they used real rattlesnakes on set. He said the animal wranglers warned him that the snakes' venomous bites would kill a human in about an hour, but "the nearest hospital's an hour and a half away." However, the snake placed in the actual coffin with Elordi was a boa constrictor with a fake rattler attached to its tail. "The snakes were rattling, which is really alarming when you're locked in a box," Elordi said in a behind-the-scenes interview for HBO. For his part, Elordi has soared to proper celebrity status since first appearing onscreen as the love-to-hate Nate, starring in movies like "Saltburn," "Wuthering Heights," and Guillermo del Toro's "Frankenstein," for which he earned his first Oscar nomination for best supporting actor. "Nate is someone who's made so many mistakes and made so many dark choices. It's cool to see it all come to what it's come to," Elordi said. "This show is a massive part of, not just my career, but my life. It's been amazing, and I'm so proud, being a part of this." The much-anticipated season three finale will air on Sunday, May 31. Although Nate's fate has been sealed, he leaves Cassie and the rest of the "Euphoria" crew in a range of precarious situations — largely at the mercy of crime lords, drug dealers, and murderous creditors impatient to collect. Given the stakes, Nate may not be the only character who meets a premature end. "They're in the real world and the consequences are real. There's no safety net," Levinson told Esquire of the season's major themes. "I like this Wild West, frontier aspect to it where you can make something of yourself, but you're going to have to live with the consequences." Read the original article on Business Insider

Sonia A. Rao / New York Times: Visually impaired Waymo users in CA say riding in a Waymo gives them a feeling of independence and spares them the discrimination they face from human drivers  —  Ruben Brunt hopped off a ferry, tapped his white cane against the sidewalk and held his phone to his ear to listen for directions.

Pope Leo XIV has just declared artificial intelligence one of the defining moral challenges of our time, in his first encyclical: a formal letter intended to guide moral, social and theological thought. Titled Magnifica Humanitas (Magnificent Humanity), it argues technology must serve humanity, rather than concentrate power or weaken human dignity. He presented it at the Vatican alongside AI developer Christopher Olah, cofounder of Anthropic, who acknowledged that companies like his need moral guidance to guard against “incentives and constraints that can sometimes conflict with doing the right thing”, the New York Times reported. “Technology is not simply a tool,” read the roughly 42,300-word open letter. “When it becomes the standard by which everything is judged, it begins to dictate what matters and what can be discarded, reducing creation to an object of exploitation and human beings to mere cogs in a system driven toward ever greater efficiency.” It warns that AI is never truly neutral, but “takes on the characteristics of those who devise, finance, regulate and use it”. And it calls for ethical oversight, social justice, protection of workers, responsible governance and peace. Automated warfare The encyclical criticises the use of AI in warfare, calling for imposing the “most rigorous ethical constraints” on weapons developed using AI. As governments invest heavily in autonomous military technologies and AI-assisted defence systems, the “growing ease” of deploying them makes war more likely and “less subject to human control”, it warns. This “violates the principle that armed force should be used only as a last resort in cases of legitimate self-defense”. The letter also criticises the growing concentration of technological power, and systems that reduce people to data or economic functions. It promotes what it calls a “civilisation of love”, centred on human dignity, solidarity, truth, compassion and the common good. Pope Leo’s response to the the AI revolution deliberately references his predecessor Pope Leo XIII’s response to the problems of the Industrial Revolution, Rerum Novarum (“Of New Things”), in 1891. Though Magnifica Humanitas was released on May 25 2026, it is symbolically dated May 15, the date of Rerum Novarum. Industrial Revolution to AI Revolution An encyclical is not an ordinary papal statement. Traditionally addressed to bishops and the wider Catholic world, it is one of the Catholic church’s most authoritative teaching documents. The pope no longer has the direct political power the papacy held in the 19th century. But papal teaching still carries moral weight across a global Catholic network of schools, universities, charities, hospitals and community organisations. The Vatican cannot regulate AI. It cannot write safety standards, police data centres, or force companies to disclose how their systems work. But it can help shape the moral terms of the debate. For more than a century, Catholic social teaching has influenced public arguments about work, inequality, poverty, human dignity and the ethical limits of economic power. Although popes issued encyclicals long before the modern era, Rerum Novarum made social encyclicals globally influential. It confronted exploitative labour conditions, widening inequality, and conflict between workers and employers. Pope Leo XIII defended workers’ rights and argued that wealth carried social responsibilities. He criticised both unrestricted capitalism and revolutionary socialism. The document influenced debates about labour rights and economic justice well beyond the church. In Australia in 1907, Justice H.B. Higgins drew on Rerum Novarum when establishing principles for a fair living wage. Pope Leo XIV’s encyclical attempts to do for the AI age what Rerum Novarum did for the industrial age: provide a moral framework for a technological transformation reshaping work, power and human relationships. Human dignity in the age of algorithms Pope Leo XIV argues human rights are not granted by governments or corporations: they arise from the intrinsic dignity of every person. Technologies should serve humanity rather than reduce people to data, economic units or optimisation problems. He builds on Pope Francis’ critique of “the tendency to let the logic of efficiency, control and profit alone shape personal, social and economic decisions”, in his 2015 encyclical. It, too, warned of the risks of technology. Pope Leo XIV argues moral responsibility can’t be transferred to automated systems, regardless of how sophisticated they become. He also rejects transhumanist ideas that human limitations should be technologically overcome, arguing vulnerability, dependence and imperfection are essential to being human. Relationships, care, solidarity and compassion are not weaknesses. “Humanity flourishes not despite limitations, but often through them.” Running throughout the encyclical is a contrast between a “culture of power” and a “civilization of love”. One treats technology primarily as a tool for domination and control. The other places human dignity, justice and care at the centre of social life. Why this matters The significance of Magnifica Humanitas lies in its ability to shape public conversation and moral imagination. Moral frameworks matter. They influence what societies fear, what they tolerate, what they defend – and what they refuse to sacrifice. Governments are investing in AI capability while still developing frameworks for transparency, accountability and safe deployment. Businesses are adopting AI tools at speed. Schools and universities are rethinking assessment, authorship and learning. Workers are being asked to adapt to systems they did not design and often cannot challenge. And citizens are increasingly governed, assessed and targeted by automated systems they may never see. Pope Leo XIV’s intervention reminds us the central question is not whether AI will be powerful: it already is. The question is whether that power will be made answerable to human dignity. The future of AI will not just be decided in laboratories, boardrooms or parliaments. It will also be decided by the moral limits societies are willing to set. Pope Leo XIV’s encyclical is an attempt to draw those limits. The authors do not work for, consult, own shares in or receive funding from any company or organisation that would benefit from this article, and have disclosed no relevant affiliations beyond their academic appointment.

<p>Dropped this morning by the Vatican: <a href="https://www.vatican.va/content/leo-xiv/en/encyclicals/documents/20260515-magnifica-humanitas.html">Magnifica Humanitas of His Holiness Pope Leo XIV on Safeguarding the Human Person in the Time of Artificial Intelligence</a>. This is a <em>very interesting</em> document. It's some of the clearest writing I've seen on the ethics of integrating AI into modern society.</p> <p>Pope Leo XIV chose the name Leo in honor of Pope Leo XIII, who is known for his 1891 <em><a href="https://en.wikipedia.org/wiki/Rerum_novarum">Rerum novarum</a></em> encyclical on "Rights and Duties of Capital and Labor".</p> <p><a href="https://www.vaticannews.va/en/church/news/2025-05/leo-xiii-s-times-and-our-own.html">This story</a> on Vatican News further clarifies the significance of that decision:</p> <blockquote> <p>Meeting with the College of Cardinals for their first formal encounter after his election, Pope Leo XIV explained part of the reason for the choice of his papal name. "There are different reasons for this," he said, before going on to explain that he chose the name Leo "mainly because Pope Leo XIII, in his historic encyclical <em><a href="https://www.vatican.va/content/leo-xiii/en/encyclicals/documents/hf_l-xiii_enc_15051891_rerum-novarum.html">Rerum novarum</a></em> addressed the social question in the context of the first great industrial revolution."</p> <p>"In our own day," he continued, "the Church offers to everyone the treasury of her social teaching in response to another industrial revolution and to developments in the field of artificial intelligence that pose new challenges for the defence of human dignity, justice, and labour."</p> </blockquote> <p>And now we get Pope Leo XIV's own encyclical on the AI revolution. There's a lot in here, but the writing style is very approachable, including to non-Catholics.</p> <h4 id="a-few-of-my-highlights">A few of my highlights</h4> <p><small>(I listened to most of the encyclical on a walk with our dog, my first time trying the <a href="https://apps.apple.com/us/app/elevenreader-read-books-aloud/id6479373050">ElevenReader iPhone app</a>. It worked very well: I pasted in a URL to the document and it read it to me in a very high quality voice, highlighting each paragraph as it went.)</small></p> <p>Here are some of my highlights. In each case below <strong>emphasis</strong> is mine.</p> <p>Here's a useful description of the interpretability problem for LLMs in section 98:</p> <blockquote> <p>First, any statement regarding AI risks becoming quickly outdated, given the remarkable pace at which these systems are developing. Second, all of us, including those who design them, possess only a limited understanding of their actual functioning. Indeed, <strong>current AI systems are more “cultivated” than “built,” for developers do not directly design every detail, but instead create a framework within which the intelligence “grows.”</strong> As a result, fundamental scientific aspects — such as the internal representations and computational processes of these systems — remain, at present, unknown.</p> </blockquote> <p>I liked section 83's description of the relationship between development and dignity:</p> <blockquote> <p>For individuals as well as for nations, development is both a duty and a right. Minimum conditions are required for enabling every person and people to flourish in accord with their dignity, without being kept in a state of dependence or excluded from access to necessary goods. Development is truly human when it places people at the center instead of the accumulation of wealth, and when it concerns peoples as well as individuals. Justice demands the recognition of the rights of society and the rights of peoples, and includes a responsibility toward future generations. <strong>Development is not truly human if it increases consumption for some while shifting costs and burdens onto others, or relegates entire regions to subordinate roles, preventing them from realizing their full potential</strong>.</p> </blockquote> <p>Baked in cultural biases and sycophancy get a mention in section 100:</p> <blockquote> <p>In personal use, three aspects in particular deserve careful consideration: the ease with which results are obtained, the impression of objectivity and the simulation of human communication. The speed and simplicity with which information, complex analyses, media content and practical assistance can be accessed undoubtedly makes life easier. Yet they can also encourage excessive reliance and the search for ready-made answers, and weaken personal creativity and judgment. <strong>The apparent objectivity of the responses and suggestions these systems provide can lead us to overlook the fact that they reflect the cultural assumptions of those who designed and trained them, with all their strengths and limitations</strong>. The artificial imitation of positive human communication — words of advice, empathy, friendship and even love — can be engaging and at times genuinely helpful. <strong>However, for less discerning users, it can also be misleading, creating the illusion of a relationship with a real personal subject</strong>. When words are simulated, they do not build genuine relationships, but only their appearance. The artificial imitation of care or support can become particularly risky when it enters contexts where real relationships and emotional bonds are lacking.</p> </blockquote> <p>101 touches on the environmental impact:</p> <blockquote> <p>Current AI systems require enormous amounts of energy and water, significantly influencing carbon dioxide emissions, and place heavy demands on natural resources. <strong>As their complexity increases, especially in the case of large language models, the need for computing power and storage capacity grows too, which requires an extensive network of machines, cables, data centers and energy-intensive infrastructure</strong>. For this reason, it is essential to develop more sustainable technological solutions that reduce environmental impact and help protect our common home.</p> </blockquote> <p>102 covers the risks of algorithmic systems making decisions that impact people's lives without "compassion, mercy, forgiveness":</p> <blockquote> <p>The use of AI is never a purely technical matter: <strong>when it enters processes that affect people’s lives, it touches on rights, opportunities, status and freedom</strong>. Important and sensitive decisions — concerning employment, credit, access to public services or even a person’s reputation — <strong>risk being fully delegated to automated systems that do not know “compassion, mercy, forgiveness, and above all, the hope that people are able to change,”</strong> and can therefore give rise to new forms of exclusion.</p> </blockquote> <p>105 emphasizes the need for human accountability in how these systems are applied:</p> <blockquote> <p>For AI to respect human dignity and truly serve the common good, responsibility must be clearly defined at every stage: <strong>from those who design and develop these systems to those who use them and rely on them for concrete decisions</strong>. In many cases, however, the internal processes leading to a result remain opaque, making it harder to assign responsibility and correct errors. <strong>This is where accountability becomes crucial: the possibility of identifying who must “account” for decisions, justify them, monitor them, and, when necessary, challenge them and remedy any harm caused</strong>.</p> </blockquote> <p>And 108 touches on the way AI amplifies the power of those with resources:</p> <blockquote> <p>In fact, as with every major technological shift, <strong>AI tends to amplify the power of those who already possess economic resources, expertise and access to data</strong>. In light of the common good and the universal destination of goods, this raises serious concerns, since small but highly influential groups can shape information and consumption patterns, influence democratic processes and steer economic dynamics to their own advantage, undermining social justice and solidarity among peoples. For this reason, it is essential that the use of AI, especially when it touches on public goods and fundamental rights, be guided by clear criteria and effective oversight, grounded in participation and subsidiarity.</p> </blockquote> <p>That same section explicitly calls out data as something that should be thought of more as a public good:</p> <blockquote> <p>[...] Moreover, <strong>ownership of data cannot be left solely in private hands</strong> but must be appropriately regulated. <strong>Data is the product of many contributors and should not be treated as something to be sold off or entrusted to a select few</strong>. It is necessary to think creatively in order to manage data as a common or shared good, in a spirit of participation, as <a href="https://www.vatican.va/content/john-paul-ii/en.html">Saint John Paul II</a> already suggested regarding collective goods.</p> </blockquote> <p>Given that Palantir is named after a <em>Lord of the Rings</em> reference, I can't help but wonder if the J.R.R. Tolkien quote from <em>The Return of the King</em> (section 213) was the Pope throwing a little shade at Peter Thiel.</p> <blockquote> <p>The twentieth-century Catholic author J.R.R. Tolkien, in the words of a protagonist in one of his novels, described our responsibility in this way: “It is not our part to master all the tides of the world, but to do what is in us for the succour of those years wherein we are set, uprooting the evil in the fields that we know, so that those who live after may have clean earth to till.” The civilization of love will not arise from a single or spectacular gesture, but from the sum total of small and steadfast acts of fidelity that serve as a bulwark against dehumanization. For this reason, it is worthwhile pausing to reflect on some aspects of how we, each in our own way, can cooperate in building the civilization of love.</p> </blockquote> <h4 id="another-2026-prediction-down">Another 2026 prediction down</h4> <p>On 6th January this year I joined the <a href="https://oxide-and-friends.transistor.fm/episodes/predictions-2026">Oxide and Friends 2026 predictions</a> podcast episode to talk about predictions for 2026, 2029 and 2032. I <a href="https://simonwillison.net/2026/Jan/8/llm-predictions-for-2026/">wrote mine up here</a>, with hindsight they weren't nearly ambitious enough - it's already undeniable that LLMs write good code, we've made huge advances in sandboxing and New Zealand kākāpō have indeed <a href="https://news.mongabay.com/short-article/2026/03/critically-endangered-kakapo-parrot-has-standout-breeding-season/">had a truly excellent breeding season</a>.</p> <p>There's one segment from the episode that I didn't bother to include in my write-up, but that I can't resist providing as a lightly-edited transcript here:</p> <blockquote> <p><strong>Bryan Cantrill:</strong> <a href="https://oxide-and-friends.transistor.fm/episodes/predictions-2026/transcript#t=37m13s">37:13</a></p> <p>I think that AI has created some real public perception problems for itself. And I think that you are gonna have one of the frontier model companies, this year, have a white paper explaining how the proliferation of AI will mean prosperity for everybody. They will be trying to make some economic argument - because this is gonna be a 2026 election issue, how we think of these things and how they are regulated and it's a big mess. There's more heat than light in this debate.</p> <p><strong>Simon Willison:</strong> <a href="https://oxide-and-friends.transistor.fm/episodes/predictions-2026/transcript#t=38m5s">38:05</a></p> <p>I'd like to tag something on to that one: I think that only works if they can sort of wash that through existing trusted experts. Sam Altman and Dario are constantly publishing essays about this stuff and nobody believes a word they say. Get Barack Obama's signature on one of these position papers and <em>maybe</em> you've got something people might start to trust a little bit.</p> <p><strong>Adam Leventhal:</strong> <a href="https://oxide-and-friends.transistor.fm/episodes/predictions-2026/transcript#t=38m27s">38:27</a></p> <p>Otherwise, it's just like "leaded gas is good for you", says Exxon.</p> <p><strong>Bryan Cantrill:</strong> <a href="https://oxide-and-friends.transistor.fm/episodes/predictions-2026/transcript#t=38m31s">38:31</a></p> <p>I mean, yeah. God. Obama... let's go with that, that's a great one because if it's like Bill Clinton everyone's gonna kind of roll their eyes, so it's gotta be someone who's got real credibility saying that this is gonna be broad-based... I'd say if they get that person to do it, it's gonna be revealed that that's also a bit crooked.</p> <p><strong>Simon Willison:</strong> <a href="https://oxide-and-friends.transistor.fm/episodes/predictions-2026/transcript#t=38m57s">38:57</a></p> <p>How about the Pope?</p> <p><strong>Bryan Cantrill:</strong> <a href="https://oxide-and-friends.transistor.fm/episodes/predictions-2026/transcript#t=39m1s">39:01</a></p> <p>The Pope is very into this stuff! That's a great prediction. We've hit pay dirt. The Pope weighing in on LLMs and their economic impact on the world.</p> <p>Simon, I'm giving you full credit if the Pope weighs in believing that this is gonna be economic devastation.</p> </blockquote> <p>My prediction here looks a whole lot less insightful given the Leo XIV/Leo XIII relationship, which I was unaware of when we recorded the episode!</p> <p>Tags: <a href="https://simonwillison.net/tags/predictions">predictions</a>, <a href="https://simonwillison.net/tags/ai">ai</a>, <a href="https://simonwillison.net/tags/kakapo">kakapo</a>, <a href="https://simonwillison.net/tags/generative-ai">generative-ai</a>, <a href="https://simonwillison.net/tags/llms">llms</a>, <a href="https://simonwillison.net/tags/bryan-cantrill">bryan-cantrill</a>, <a href="https://simonwillison.net/tags/ai-ethics">ai-ethics</a></p>

Bryan Hecht / The Boston Globe: Massachusetts formally recognizes the App Drivers Union, which says it represents ~70,000 workers and is the first state-certified rideshare union in the US  —  Uber and Lyft drivers in Massachusetts have officially unionized, a first-in-the-nation move that now sets up what could be the final stage in a years-long organizing effort.

<p><img src="https://static.inaturalist.org/photos/666934915/large.jpg" alt="California Brown Pelican"></p><p><img src="https://static.inaturalist.org/photos/666934945/large.jpg" alt="California Brown Pelican"></p><p><img src="https://static.inaturalist.org/photos/666934484/large.jpg" alt="Snowy Egret"></p><p><img src="https://static.inaturalist.org/photos/666935110/large.jpg" alt="California Sea Lion"></p><p><img src="https://static.inaturalist.org/photos/666935468/large.jpg" alt="Harbor Seal"></p><p>California Brown Pelican, Snowy Egret, California Sea Lion, Harbor Seal, in San Mateo County, CA, US</p><p>We took our new <a href="https://www.orukayak.com/">folding kayak</a> out in the harbor and saw sea lions and harbor seals chilling on the docks.</p>

Adi Robertson / The Verge: The FTC settles with Cox, MindSift, and 1010 Digital Works for $930K over claims they falsely said they could use phone mics to spy on users for ad targeting  —  More specifically, it was fined for allegedly lying by claiming that it could. … An exceptionally weird controversy has come …

Discussion | Link

In an age of AI, Pope looks for "artisans of hope."

Discussion | Link

Lakshmi Varanasi / Business Insider: X says it is cracking down on large accounts that have been gaming its revenue-sharing program by “programmatically reuploading content from smaller accounts”  —  X is introducing new strategies to rein in its copycat economy.  —  Elon Musk's social media platform is now cracking …

Reuters: Iranian state media reports that President Masoud Pezeshkian has issued an order to reopen international internet access after a near-90-day blackout  —  Iran's President Masoud Pezeshkian has issued an order to reopen international internet access, Iranian state media reported on Monday …

Marten Newhall/Unsplash Global society makes billions of images and uploads hundreds of thousands of hours of video on the internet every day. The problem is, some of this content is misleading or downright wrong. And when it’s in visual form, it can be particularly convincing. Take the Met Gala that happened earlier this month in New York. While photographers snapped photos of Rhianna, Beyoncé and Nicole Kidman as they strutted their stuff, others saw “photos” of celebrities, such as Rosalía, Lady Gaga and Jacob Elordi, who were actually elsewhere (the images in the below Instagram carousel are AI generated). While this type of AI slop might seem harmless and can be easily verified, other “media fakery” is becoming far more problematic and demands more robust techniques to verify. Traditional verification techniques are falling short as AI becomes increasingly convincing and the line between authentic and synthetic blurs. This is true across all content, from still images to moving ones and audio deepfakes. The volume of content and the speed at which it travels doesn’t help. It also doesn’t help that fact-checking can take hours or days while fakes can be created in seconds. First, equip yourself Guides on detecting AI-generated content suggest multiple strategies and acknowledge there are no perfect solutions. But there are helpful things you can do. Familiarise yourself with examples of fakes and study how they were fact-checked. This helps you understand what is possible and learn how fact-checkers sort real from fake. Look deeply. Zoom in. Pause the content or watch it frame-by-frame. Inspect the small details. Look out for inconsistencies, textures that are flat when they shouldn’t be, or patterns that are too perfect or are inexplicably off. Does the location shown match with where the scene is purported to be? Do shadows fall naturally and do lines follow the rules of perspective? Look widely. Are you familiar with the source? What else does it publish and how long has it been around? What do other trusted sources say? How does this depiction compare to others that are available? Or if there aren’t others available, should that give you pause? Then, apply your learnings Let’s take an example and work through it together. This Facebook reel, posted by an account called “Real Talk Hub”, purports to show migrants being stopped and returned by Australian police at an airport. Before getting too granular, let’s take stock of the opening image. The video uses scale to show what appears to be a long stream of passengers. Some are moving toward and some are moving away from a plane. It is difficult to identify specifics in the video. The superimposed text blocks almost all of the horizon line. Shallow depth of field makes aspects in the distance blurry and hard to discern. Many of the passengers have darker skin and are visually coded as “other”. They interact with a light-skinned police officer who takes notes on a clipboard. The vertical video is framed carefully to not reveal identifiers like the name of the airline that seems to start with the letter “P”. This makes it difficult to search the airline’s name and whether credible sources corroborate the story that’s told. Even though the people and scenes look realistic at first glance, the video’s integrity unravels when we slow down and look closer. People in the passenger line morph and transform. The officer is able to single-handedly remove the paper from the clipboard and it appears to inexplicably leave white strips behind. The police vests look different to images you can find in verified media photos of the Australian Federal Police. Taken together, all these clues suggest the video is AI-generated. The paper on the clipboard moves in an unrealistic way, and the police vest is not accurate. Real Talk Hub/Facebook Think like a fact-checker Many AI-generated videos can trick you and create a very compelling narrative. So, fact-checkers have developed triangulated methodologies that examine elements beyond just what you see in the video. One way to do this is to systematically check contextual factors – the other things surrounding the content. Our team’s research has found professional fact-checkers usually pay attention to the type of social media accounts or websites distributing suspicious media. For this AAP verification on a video about banning dogs on the beach, it was crucial to inspect the user’s activity and posting patterns. In addition to visual anomalies, the fact-checkers also found an invisible watermark that helped them determine the content was AI-generated. Other things to check are how long a social media account has been operating, how often the social media account posts, and whether the account is transparent about its use of AI. These aren’t fool-proof indicators of authenticity, though. The migrant example above comes from an account that is about five years old. It also comes from a “verified” account, which might make it feel more credible. But both Facebook and X now let users pay for this verification. Overall, when it comes to suspect images or video, don’t just look deeply. Also look widely. AI-generated content can increasingly fool our eyes, so you also have to look beyond what’s in the video. Taking a mixed-methods approach that considers visual and contextual clues can help. By training your ability to think like a fact-checker, you can stay safer online. We teach young people to write. In the age of AI, we must teach them how to see Silvia Montaña-Niño is also associate investigator of the ARC Centre of Excellence for Automated Decision-Making & Society and the Fact Check Research Team at this centre. T.J. Thomson receives funding from the Australian Research Council. He is an affiliate with the ARC Centre of Excellence for Automated Decision Making & Society.